Z6 ii and Z7 ii

Pistnbroke

If the camera wont make the lens focus on the bird and keep track of it (Z7) then that part of the system is a failure . You dont need experience or intelligence to know that and your memory tells you how good the D850 is.

photobunny

tc88 said:

WestEndFoto said:

Nobody was doomed when they got leapfrogged. It is easy to forget that if you got leapfrogged, you probably leapfrogged someone on the next cycle.

But there are lots of people with either little experience, short memories or a low IQ that don't see this.

Well film cameras are very different from digital cameras in terms of expertise required. Has Canon leapfrogged Nikon in image quality in the last 10 years? And is 10 years a long time?

To think Nikon will leapfrog Sony in AF in the next iteration of mirrorless is pure wishful thing, because in mirrorless, the ceiling for AF is the sensor because it determines how fast the data can be pumped out.

To think Nikon can somehow produce a better sensor on Sony fabs than Sony can produce internally is pure nonsense. Whatever technology Nikon is allowed to use, Sony will have tested first. So unless Nikon finds another fab (say Towerjazz) and that fab somehow manage to produce better sensors than Sony, there is no chance Nikon's sensor will be better. And in semiconductor manufacturing, while leapfrogging happens, it's very rare.

The AF on mirrorless is all CPU bound. Nikon are behind because their CPU is only up to task for the traditional modes and not the smart eye AF modes. Canon are ahead of everyone because the Digic X is a much faster chip. The sensors can remain exactly the same for another decade and you will still see markable improvements in AF as long as CPU's continue to get faster.

Nikon could take a nearly off the shelf ARM CPU to TMSC 5nm process and suddenly they have the most responsive AF on the market. What they did wrong was rather arrogantly using a chip that was more than good enough for a DSLR and then ask it to perform all the other duties a mirrorless camera also needs. Cricky you can turn various CPU intensive things off and suddenly the AF on the Z6 gets that wee bit faster.

snakebunk

I don't think the differences in our views adhere to intelligence or exprience as much as different use cases.

For me the lens choice is not too hard. You can choose any long prime lens from Nikon, Canon, Sony or Sigma, and I expect them to be fairly equal. I am still happy with my Sigma 500/4 and don't feel any urgent need for improvement. But all long lenses can be lighter or have a faster and more accurate af, better stabilization etc. and there is a healthy competition.

Now there is a rumor of a 400/2.8 (I was happy to see that) for the Z mount, but there are rumors of five long prime lenses for the RF mount. I am not worried that Nikon will dissapear any time soon but I am worried that Nikon will not be able to keep up with Canon (and maybe Sony) when it comes to long lenses.

There are lots of interesting camera characteristics, like af, frame rate, iso, resolution, ergonomics and build quality. And for me the choice of camera is at least as interesting as the lens.

The rumors of the new Z6, Z7 and the Z9 are very interesting, but I need to see tests that compare the af to the competition. If the new cameras focus as good or better than Canon, Sony and DSLRs, and there is also a 600/4 on the horizon, it could be bird photography success for Nikon. I think. But they need to step up.

PS. I have never seen any distancing between bird photographers because of the brands. It really sounds rediculous. There are so many aspects that are more important than equipment.

photobunny

The 5 RF super teles incoming are all starting to whisper to me. Nikon has the 500mm f/5.6 PF but the RF super tele lenses could all be DO's and then you have little fat f/4 lenses that you can hand hold (within reason, but my main lens was a 300mm f/2.8 from 1988 that I handheld on a 5DII and 7D without issue).

tc88

photobunny said:

The AF on mirrorless is all CPU bound.

That's just not true. The reason A9 can AF so well is because it can refresh up to 60fps for AF. Other mirrorless sensors are no where close to that. If your sensor can only pump out 10fps for AF, then the accuracy of the movement prediction is not going to be as good. It doesn't matter how much processing power you throw at it. There is just no timely data to process.

ggbutcher

tc88 said:

photobunny said:

The AF on mirrorless is all CPU bound.

That's just not true. The reason A9 can AF so well is because it can refresh up to 60fps for AF. Other mirrorless sensors are no where close to that. If your sensor can only pump out 10fps for AF, then the accuracy of the movement prediction is not going to be as good. It doesn't matter how much processing power you throw at it. There is just no timely data to process.

The sensors I work with (not consumer camera ones) rely on the connected CPU to clock the data out of the ADCs. If that's also the case in our camera sensors, then 'CPU-bound' would be an appropriate assertion. Someone more familiar with camera architecture can pipe in...

tc88

ggbutcher said:

The sensors I work with (not consumer camera ones) rely on the connected CPU to clock the data out of the ADCs. If that's also the case in our camera sensors, then 'CPU-bound' would be an appropriate assertion. Someone more familiar with camera architecture can pipe in...

Of course everything input to the CPU is synchonized to the CPU clock. But that will be at least mhz range. And when the CPU is fast enough, it doesn't mean there is always new data every clock cycle if the ADC can't keep up supplying relevant new data.

So the CPU clock speed has no relevance to the sensor fps here. It's like on a computer, the memory and CPU speeds are completely separate, and increasing the CPU speed does not improve memory bandwidth or latency.

PB_PM

ggbutcher said:

tc88 said:

photobunny said:

The AF on mirrorless is all CPU bound.

That's just not true. The reason A9 can AF so well is because it can refresh up to 60fps for AF. Other mirrorless sensors are no where close to that. If your sensor can only pump out 10fps for AF, then the accuracy of the movement prediction is not going to be as good. It doesn't matter how much processing power you throw at it. There is just no timely data to process.

The sensors I work with (not consumer camera ones) rely on the connected CPU to clock the data out of the ADCs. If that's also the case in our camera sensors, then 'CPU-bound' would be an appropriate assertion. Someone more familiar with camera architecture can pipe in...

Indeed, this is why Nikon is going with dual processors on the the Z7II/6II. Sensor readout is an issue for the contrast detect phase of focus, but not the phase detect phase of AF. That is why the phase detect was added to mirrorless cameras, it helps reduce the lag in AF speed from the sensor refresh rate. So the reality is that it is a bit of both. In either case, both the sensor readout and phase detect aspects are affected by CPU speed at the end of the day. The sensor readout because no matter how fast the sensor reads out, the CPU has to be able to process that information quickly to make the focusing algorithms work faster.

WestEndFoto

PB_PM is correct. Nikon executives clearly stated that further AF improvements are not possible without a hardware upgrade in the DPReview interview. The only hardware upgrade coming Wednesday is a dual processor. The sensor is the same. Nikon is stating AF will be improved. Bob is your uncle.

The same thing happened when Nikon updated the D4 to the D4s. Nikon updated from Expeed 3 to Expeed 4. I remember watching YouTube videos trying to discern whether af was better on the Nikon D4 or it’s Canon competitor. With the D4s and Expeed 4, Nikon never looked back.

WestEndFoto

PS:
Being second to Nikon on af and dynamic range did not impact Canon’s market share much. Nikon will be fine if it is not first on one or two features, except in the minds of some YouTube influencers.

Yes YouTube influencers, I am talking to you. Except for a few babies not being born because a few measurbators instead soiled the gear you convinced them to buy, you guys are irrelevant.

WestEndFoto

One more PS:
I do think sensor readspeed is important at some point. But according to Nikon, the current bottleneck on the Z6 and Z7 is the processor, not the sensor.

Perhaps with the dual Expeed, it will be the sensor. Or perhaps not until Expeed 6?

tc88

WestEndFoto said:

One more PS:
I do think sensor readspeed is important at some point. But according to Nikon, the current bottleneck on the Z6 and Z7 is the processor, not the sensor.

Yes, that looks to be true. The current models probably are already limited by the processing power. But there is question that even with full supporting processing power, how the AF on those models really performs. Note they didn't change the sensors.

In reality, those chips, while having some special functionality, are really cheap, low end chips. It's very easy to bring up the performance of those. Nikon was either (1) being extra cheap, or (2) lacking even the engineering resource to incorporate those in the first iteration.

Now just to be clear, the sensor has its own operating frequency and architecture. So clocking the CPU or improving the performance beyond a certain point (which is easy to do) will do not a thing on the overall performance. That's why sensor will always be the showstopper technology wise, not the CPU.

snakebunk

When it comes to af on modern cameras I think there is also a lot of software engineering involved.

tc88

snakebunk said:

When it comes to af on modern cameras I think there is also a lot of software engineering involved.

Of course, if we are talking about those fancy AF modes.

But if the benchmark is D850 in phase detect, then I suppose those software have already been written. So in theory Z7 should be able to meet it if it has comparable processing unit. The reason it can't meet the D850 phase detect performance is because it doesn't have the dedicated AF sensor which means (1) it can't get the phase data out in a high enough frame rate, and (2) it can't have large enough phase detect sites. Both of those are only related to the sensor.

PB_PM

tc88 said:

In reality, those chips, while having some special functionality, are really cheap, low end chips. It's very easy to bring up the performance of those. Nikon was either (1) being extra cheap, or (2) lacking even the engineering resource to incorporate those in the first iteration.

It's not just the price in terms cost of the processor, but also in battery life. Mobile CPU power draw has greatly improved in the last two years (in terms of performance/watt), so Nikon can now add dual processors without compromising battery life. It was likely a compromise they had to make at the time.

tc88

PB_PM said:

It's not just the price in terms cost of the processor, but also in battery life. Mobile CPU power draw has greatly improved in the last two years (in terms of performance/watt), so Nikon can now add dual processors without compromising battery life. It was likely a compromise they had to make at the time.

That's a valid point. But if I were the manager, I would add it if it really helps the performance significanly. You don't need to use all the performance. It can certainly be enabled to be on demand usage.

Finally are you sure the chips used in Z7ii will be a newer version? I highly doubt it since it will have to be made on a different process node to get the power saving which costs millions of dollars to create all the new masks. I would imagine it will be given a newer name too.

PB_PM

I suspect they are moving to whatever the D6 is using. It might not be a new architecture, but it could still be a die shrink. Expeed is an in house chip design, but someone else has the fabs (Nikon doesn't have any that I know of), so whatever processor node that fab is using is what Nikon will get. I suspect they are using TSMC, which has moved from a 12nm to 7nm process in the last few years, which would net Nikon a good bump in performance per watt.

WestEndFoto

tc88 said:

PB_PM said:

It's not just the price in terms cost of the processor, but also in battery life. Mobile CPU power draw has greatly improved in the last two years (in terms of performance/watt), so Nikon can now add dual processors without compromising battery life. It was likely a compromise they had to make at the time.

That's a valid point. But if I were the manager, I would add it if it really helps the performance significanly. You don't need to use all the performance. It can certainly be enabled to be on demand usage.

Finally are you sure the chips used in Z7ii will be a newer version? I highly doubt it since it will have to be made on a different process node to get the power saving which costs millions of dollars to create all the new masks. I would imagine it will be given a newer name too.

My guess is that they are not an updated Expeed. But the time is about right for Expeed 7 to come out. So we might be surprised. I would then expect things to get really interesting AF wise.

WestEndFoto

tc88 said:

WestEndFoto said:

One more PS:
I do think sensor readspeed is important at some point. But according to Nikon, the current bottleneck on the Z6 and Z7 is the processor, not the sensor.

Yes, that looks to be true. The current models probably are already limited by the processing power. But there is question that even with full supporting processing power, how the AF on those models really performs. Note they didn't change the sensors.

In reality, those chips, while having some special functionality, are really cheap, low end chips. It's very easy to bring up the performance of those. Nikon was either (1) being extra cheap, or (2) lacking even the engineering resource to incorporate those in the first iteration.

Now just to be clear, the sensor has its own operating frequency and architecture. So clocking the CPU or improving the performance beyond a certain point (which is easy to do) will do not a thing on the overall performance. That's why sensor will always be the showstopper technology wise, not the CPU.

I am certainly looking forward to faster sensors and a global shutter implementation could make things interesting.

tc88

WestEndFoto said:

My guess is that they are not an updated Expeed. But the time is about right for Expeed 7 to come out. So we might be surprised. I would then expect things to get really interesting AF wise.

Yes, I would expect Expeed 7 should allow Z7ii to catch up to the other mirrorless (other than A9). But considering Expeed 5 was released Jan 2016 and Expeed 6 August 2018, I would guess Expeed 7 will have to come with Z8/Z9. Still Expeed 6 x 2 should still be powerful enough that the processing power should not be the limiting factor.

PB_PM said:

I suspect they are moving to whatever the D6 is using. It might not be a new architecture, but it could still be a die shrink. Expeed is an in house chip design, but someone else has the fabs (Nikon doesn't have any that I know of), so whatever processor node that fab is using is what Nikon will get. I suspect they are using TSMC, which has moved from a 12nm to 7nm process in the last few years, which would net Nikon a good bump in performance per watt.

Very few people know the details of those architectures and die shrink certainly qualify as a new product name in pretty much every company. So we will find out soon if it's just same chip x 2 or not.

WestEndFoto said:

I am certainly looking forward to faster sensors and a global shutter implementation could make things interesting.

Those are the technologies that will finally allow mirrorless to overtake DSLR. Unfortunately those are all chip technologies and Nikon will not be the first to those.

WestEndFoto

Why so certain re: Nikon being behind on chip technologies?

tc88

@WestEndFoto, I'm referring to the sensor chip technology, not the generic technology that manufactures DSP or CPU. Sony is the leader in sensor chip technology, and as long as Nikon has to manufacture through Sony fabs, there is no way Nikon can lead.

Because whatever chip technology it's allowed to use, it has to be given by Sony after tested by Sony in its fabs. So unless another company like TowerJazz can surpass Sony someday which is a slim chance currently, there is just little chance.

rmp

I do not know he internal software/hardware architecture of a camera, but reading what this group is saying makes me want to but a ZII just to play with the AF.

photobunny

rmp said:

I do not know he internal software/hardware architecture of a camera, but reading what this group is saying makes me want to but a ZII just to play with the AF.

I am looking forward to trying the Z6 and Z6II side by side to see how they perform on AF... though I might go for the Z7II if its bust has a 'real' 10 FPS mode and not the weird slideshow mode in the Z6 H+ mode.

PB_PM

Nikon's edg

tc88 said:

Very few people know the details of those architectures and die shrink certainly qualify as a new product name in pretty much every company. So we will find out soon if it's just same chip x 2 or not.

Nikon already uses the same name for various versions of the same generation of chips (same architecture is about all they share). Do really believe that the D5s, D850 and the D7500 use the exact same chip? They don’t, even the part numbers are different, but they are all labeled Expeed 5.
The Expeed 5A was a revision of it as well. Officially it was only shipped in a Nikon 1 camera though.